# Copyright 2018 Amazon.com, Inc. or its affiliates. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License").
# You may not use this file except in compliance with the License.
# A copy of the License is located at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# or in the "license" file accompanying this file. This file is distributed
# on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either
# express or implied. See the License for the specific language governing
# permissions and limitations under the License.
import math
from typing import Optional, Tuple
import numpy as np
from gluonts.core.component import validated
from gluonts.mx import Tensor
from gluonts.mx.linalg_util import lower_triangular_ones
from .distribution import Distribution, _sample_multiple, getF
from .distribution_output import DistributionOutput
[docs]class MultivariateGaussian(Distribution):
r"""
Multivariate Gaussian distribution, specified by the mean vector and the
Cholesky factor of its covariance matrix.
Parameters
----------
mu
mean vector, of shape (..., d)
L
Lower triangular Cholesky factor of covariance matrix, of shape
(..., d, d)
F
A module that can either refer to the Symbol API or the NDArray
API in MXNet
"""
is_reparameterizable = True
@validated()
def __init__(self, mu: Tensor, L: Tensor, F=None) -> None:
self.mu = mu
self.L = L
@property
def F(self):
return getF(self.mu)
def __getitem__(self, item):
raise NotImplementedError()
@property
def batch_shape(self) -> Tuple:
return self.mu.shape[:-1]
@property
def event_shape(self) -> Tuple:
return self.mu.shape[-1:]
@property
def event_dim(self) -> int:
return 1
[docs] def log_prob(self, x: Tensor) -> Tensor:
# todo add an option to compute loss on diagonal covariance only tosave
# time
F = self.F
# remark we compute d from the tensor but we could ask it to the user
# alternatively
d = F.ones_like(self.mu).sum(axis=-1).max()
residual = (x - self.mu).expand_dims(axis=-1)
# L^{-1} * (x - mu)
L_inv_times_residual = F.linalg_trsm(self.L, residual)
ll = (
F.broadcast_sub(
-d / 2 * math.log(2 * math.pi), F.linalg_sumlogdiag(self.L)
)
- 1
/ 2
* F.linalg_syrk(L_inv_times_residual, transpose=True).squeeze()
)
return ll
@property
def mean(self) -> Tensor:
return self.mu
@property
def variance(self) -> Tensor:
return self.F.linalg_gemm2(self.L, self.L, transpose_b=True)
[docs] def sample_rep(
self, num_samples: Optional[int] = None, dtype=np.float32
) -> Tensor:
r"""
Draw samples from the multivariate Gaussian distributions. Internally,
Cholesky factorization of the covariance matrix is used:
sample = L v + mu,
where L is the Cholesky factor, v is a standard normal sample.
Parameters
----------
num_samples
Number of samples to be drawn.
dtype
Data-type of the samples.
Returns
-------
Tensor
Tensor with shape (num_samples, ..., d).
"""
def s(mu: Tensor, L: Tensor) -> Tensor:
F = self.F
samples_std_normal = F.sample_normal(
mu=F.zeros_like(mu),
sigma=F.ones_like(mu),
dtype=dtype,
).expand_dims(axis=-1)
samples = (
F.linalg_gemm2(L, samples_std_normal).squeeze(axis=-1) + mu
)
return samples
return _sample_multiple(
s, mu=self.mu, L=self.L, num_samples=num_samples
)
[docs]class MultivariateGaussianOutput(DistributionOutput):
@validated()
def __init__(self, dim: int) -> None:
self.args_dim = {"mu": dim, "Sigma": dim * dim}
self.distr_cls = MultivariateGaussian
self.dim = dim
self.mask = None
[docs] def domain_map(self, F, mu_vector, L_vector):
# apply softplus to the diagonal of L and mask upper coefficient to
# make it lower-triangular diagonal matrix whose elements are diagonal
# elements of L mapped through a softplus
d = self.dim
# reshape from vector form (..., d * d) to matrix form(..., d, d)
L_matrix = L_vector.reshape((-2, d, d, -4), reverse=1)
L_diag = F.broadcast_mul(
F.Activation(
F.broadcast_mul(L_matrix, F.eye(d)), act_type="softrelu"
),
F.eye(d),
)
mask = lower_triangular_ones(F, d, offset=1)
L_low = F.broadcast_mul(L_matrix, mask)
return mu_vector, L_diag + L_low
@property
def event_shape(self) -> Tuple:
return (self.dim,)